FasTrak SoftWorks

Proposal 1: Sovereignty‑Preserving, ICAO‑Style UN Organization for Capacity Building for Globally Interoperable AI Safety Rules Recommendation for the UN Global Dialogue on AI Governance to call for the establishment of a permanent, multistakeholder body modeled after the International Civil Aviation Organization (ICAO). This body shall develop voluntary, sovereignty‑preserving technical standards consistent with international human rights law, facilitate capacity building for developing nations, and promote interoperability among national AI governance frameworks. This body also serves as the institutional host for related guidance instruments. Proposal 2: AI Product Digital Passport Recommendation for the UN Global Dialogue to endorse the creation of an AI Product Digital Passport as a standardized, machine‑readable digital document that traces an AI model or system across its entire lifecycle, from development through deployment, enabling supply chain accountability and human oversight while affording jurisdiction‑specific and privacy-enabled control over its use. Developed under the ICAO‑style body (Proposal 1) to ensure global compatibility. Proposal 3: Global AI Energy and Sustainability Dashboard The UN Global Dialogue calls for the creation of a global public dashboard that tracks energy, water, and critical minerals consumption attributable to AI development and deployment, recognizing the rights of communities affected by extraction and hosting. The dashboard shall provide transparency for the public, developers, deployers, policymakers, and governments regarding the planetary‑scale costs of AI. The dashboard aggregates anonymized, voluntarily reported data to project standardized metrics that are meaningful and accessible to all stakeholders. Proposal 4: Risk Reduction through Supply Chain Transparency of Model Provenance Tracing The UN Global Dialogue endorses voluntary, non‑binding model provenance tracing as a risk reduction mechanism. Based on risk classification, AI developers and deployers are encouraged to implement verifiable AI model provenance computing (akin to DNA tests) from first release through each fine-tuning (adaptation). AI model provenance checking during inference, can confirm that the model in use matches its declared provenance and has not been swapped or tampered with. Proposal 5: Digital Embassy Guidance for Sovereignty‑Preserving, Non‑binding Rules for Bilateral Digital Infrastructure Agreements The UN Global Dialogue calls for a UN‑based initiative to draft non‑binding guidance rules for bilateral digital embassy agreements. These rules shall guide countries without sufficient AI compute infrastructure (e.g., Maldives) and hosting countries in establishing trusted, cross‑border digital relationships, while fully respecting national sovereignty and international humanitarian law.

• Safe, secure and trustworthy AI
• Interoperability of governance approaches
• AI capacity-building
• Transparency, accountability, and human oversight

AI brings a novel risk profile characterized by unprecedented speed of innovation, cross-border scaling, and value chain dependencies that create emergent failures at scale. We observe three main gaps in AI Governance today: first, is a lack of shared accountability infrastructure across the AI supply chain; second, governance that exists in documents but not in system execution; and a lack of global process standards. This profile and observed gaps drive risks to fundamental rights, public safety, national security, market integrity, and planetary sustainability. We offer that AI Governance is not a structured problem that can be 'solved' through decomposition into subproblems. Rather, it is an unstructured problem that must be continuously managed-much like financial fraud. Even with Basel Accords, fraud persists. We propose that a strategically sound goal for global AI Governance is proportionate, verifiable, and continuous management through robust international infrastructure. The global AI ecosystem is progressively bifurcating into clean AI: regulated, trusted, and economically flourishing, and dark AI: unsafe and unaccountable, deliberately beyond reach - exactly as the internet today is divided into the open web and the dark web. That bifurcation will not be reversible. Arriving at an international architecture to sustain the clean AI zone is urgent. The dark zone is actively being built through permissionless access to AI models of increasing capability and increasingly lower technical barriers to deployment. We propose a multi-prong risk management strategy for global AI Governance policy development: (1) Align with existing human rights-based policy frameworks and initiatives as a foundation for continued development, namely, OECD AI Principles, UNESCO Ethical AI Recommendation, Hiroshima Process, Council of Europe AI Treaty, and the EU AI Act. (2) Enhance supply chain visibility throughout the lifecycle of AI systems. (3) Build cross-border collaboration initiatives towards sovereignty-preserving global coordination. The proposals presented here, anchored to the strategy, advance safe and trustworthy AI through interoperable infrastructure (Cluster 3), bridge capacity divides (Cluster 2), surface AI's broader implications (Cluster 1), and operationalize transparency and accountability consistent with international human rights law (Cluster 4).